Use of Extrusion–Spheronization to Develop

Controlled-Release Dosage
Forms for Diltiazem
Hydrochloride
Anurag Sood, Yasvanth Ashokraj, and Ramesh Panchagnula*

D
NATIONAL INSTITUTE OF PHARMACEUTICAL
uring the past few decades, various types of oral
controlled-release (CR) formulations have been devel-
oped to improve the clinical efficacy of drugs having
short half lives as well as to increase patient compliance
EDUCATION AND RESEARCH

(1, 2). These formulations are designed to deliver drugs at a pre-

determined rate over a wide range of conditions and durations
of therapeutic treatments. Matrix-based systems, in which the
drug is dispersed as a fine powder from a matrix of polymeric-
and non-polymeric release modifying agents, are preferred CR
drug delivery systems because they are easy to manufacture (3).
The process of extrusion–spheronization (ES), introduced
by Reynolds (4) and Conine and Hadley (5), has become the
method of choice for developing multiunit, pellet-based dosage
forms. ES offers the technological advantage of providing multi-
The authors explore the possibility of using particulates with a spherical shape, good flow properties, low
friabilities, and uniform packing characteristics. In addition to
extrusion–spheronization to develop a matrix-
high drug load, the pellet offers the advantage of achieving mod-
based, controlled-release formulation for ified drug release by incorporating release modifying agents
diltiazem hydrochloride, a highly water-soluble such as ethylcellulose, acrylic polymers, chitosan, and glyceryl
drug. Various approaches involving modification monostearate in the formulation (6–9). However, the number
of the formulation composition or process were of formulation aids and release modifying excipients available
adopted to achieve the target drug-release for ES is very limited because of the properties needed for their
wet mass. Pellets can be compressed to yield a single unit dosage
profile.
form or can be coated to deliver a drug at a desired rate at the
same site or various sites within the gastrointestinal tract.
The application of ES for the development of CR formula-
tions for highly water-soluble drugs without involving a coat-
ing step in the manufacturing operation is an area of interest
because the methodology combines the advantages of a CR ma-
trix system with an ES technology. Hence, the objective of the
Anurag Sood is a senior research
scientist at Ranbaxy Research Laboratories, present study was to use ES to develop a matrix-based CR for-
(Haryana, India). Yasvanth Ashokraj is a mulation for diltiazem hydrochloride (DLT), a highly water-
doctoral research scholar and Ramesh soluble Class I drug that releases the drug at a predetermined
Panchagnula is a professor in the rate on the basis of pharmacokinetic principles. The drug has a
department of pharmaceutics at the National
saturation solubility of 590 mg/mL in phosphate buffer pH
Institute of Pharmaceutical Education and
Research, Sector 67, S.A.S. Nagar, Punjab, 7.4 (10). Various approaches involving the modification of the
160 062, India, tel.
91 172 2214682, formulation composition or process were adopted to achieve
fax
91 172 2214692, panchagnula@ the target drug-release profile. To test the effectiveness of using
yahoo.com various excipients for successful ES, batch yields were determined
*To whom all correspondence should be addressed. and pellets were evaluated for physical characteristics such as
particle size and distribution, crushing strength, and density.
62 Pharmaceutical Technology APRIL 2004 www.phar mtech.com
Excipients were screened for compatibility with DLT before
their use. Drug-release mechanisms and kinetics were eluci-
dated, and the stability of selected formulations under acceler-
ated temperature and humidity conditions were established.
Materials and methods
Materials. DLT (batch no. 0350200) was obtained from Chemi-
nor Drugs Ltd. (India). The following chemicals were used in
the formulation development:
● spheronization aid (SPA): microcrystalline cellulose (MCC),
PH 101
● low–melting point hydrophobic agents: magnesium stearate,
glyceryl monostearate, glyceryl palmitostearate, and bees wax
● water-swellable polymers: hydroxypropyl methylcellulose
(HPMC), low-substituted hydroxypropyl cellulose (L-HPC),
and hydroxypropyl cellulose
● water-insoluble excipients: ethylcellulose
● binders: HPMC, polyvinyl pyrolidone (PVP), and Eudragit
NE 30 D.
HPMC, MCC, magnesium stearate, and PVP were obtained
from Panacea Biotech Ltd. (India). Glyceryl monostearate, ethyl-
cellulose, lactose, and bees wax were purchased locally. Glyc-
eryl palmitostearate and Eudragit NE 30D were obtained from
Gattefosse (France) and Rohm GmbH (Germany), respectively.
L-HPC 11 and L-HPC 20 were obtained from Shin-Etsu (Japan).
All materials were used as received.
Methods. Theoretical design of controlled drug-release profile. Using the
superposition method described by Ritschel, a theoretical con-
trolled drug-release profile for DLT was developed on the basis
of desirable target blood concentrations and pharmacokinetic
properties of the drug (11). The steady-state drug concentra-
tion in the blood was predicted by using the same method. The
first-order drug-release rate (R0) was chosen by calculating the
target release profile for dosage forms that would release the
drug for a period of time (tDEL) shorter than the selected dos-
ing interval ( = 12 h).
Drug–excipient compatibility studies. Isothermal stress testing was
performed on binary drug–excipient mixtures by exposing the
mixtures to elevated temperatures and moisture levels to ac-
celerate drug–excipient interaction, if any. Mixtures were ana-
lyzed using differential scanning calorimetry (DSC; Mettler-
Toledo DSC 821e, Switzerland) and fourier transform infrared
spectroscopy (FT-IR; Impact 410, Nicolet, Madison, WI).
Preparation of pellets. MCC is the most frequently used excipi-
ent to aid aqueous ES because it forms a plastic and cohesive
mass upon wetting, which are the most desired characteristics
for successful ES (12, 13). Hence, a basic pellet formulation con-
sisting of the drug and MCC was used. Other excipients were
incorporated in various combinations and proportions to mod-
ify the drug-release characteristics. The ES technique—includ-
ing the uses of a rotating roller extruder (model 10, Caleva, UK)
with a standard 1-mm screen and a spheronizer (model 120)
with a 120-mm diameter (33 mm2 pitch, 1-mm depth)—was
used to prepare 50-g batches of pellets.
Weighed quantities of the raw materials passed through
British standard sieves (BSS) no. 80 (Scientific Engineering
Corp., India) and were dry mixed in a polyethylene bag for
64 Pharmaceutical Technology APRIL 2004
10–15 min. The mixing time was optimized by content unifor-
mity testing in initial batches (data not shown). The mixture
was granulated using the granulation fluid to achieve the ap-
propriate level of moisture content for extrusion and spher-
onization. The moisture content of the wet mass was deter-
mined using an infrared moisture balance (PM480 Delta Range
integrated with an LP16 IR dryer and an LC45 printer, Mettler-
Toledo) and reported as the percent weight loss of initial wet
weight. The wet granulation mixture was extruded, and the ex-
trudate was then immediately spheronized at a speed of 2950
rpm for 5 min to yield spherical pellets. The pellets were dried
in glass-lined trays in a forced-circulation, hot-air oven (Narang
Scientific Instruments, India) at 40 C for 10–12 h. The dried
pellets were then filled and stored in screw-capped, high-
density polyethylene (HDPE) bottles.
Some pellet formulations were prepared by dispersing the
drug in molten, low–melting point hydrophobic agents as re-
lease retardants by heating them to their melting temperatures
and then the drug was dispersed in it. The mixture then was
cooled to room temperature under constant stirring to produce
a solid mass that was either crushed by hand or milled with
laboratory-scale multimill equipment (Gansons Ltd., India) to
generate uniform sized powder and granules (passed through
BSS sieve no. 30). To avoid product loss during preparation,
low–melting point hydrophobic agents were prepared in slight
excess so that the required quantity of powder (passed through
BSS no. 80 mesh) was weighed and used for ES following the
described procedure.
Compression of DLT pellets (size fraction passed through BSS
no. 22 and retained on BSS no. 44 sieve; i.e., 355–1000 µm size
range, unless specified) was performed using a single-punch
tablet press fitted with either 9- or 12.7-mm circular flat-faced
punches. Weighed quantities of pellets, premixed with 0.2% w/w
aerosil, were filled manually into the die and compacted man-
ually. Tablets were stored in tightly closed HDPE bottles. Selected
formulations were compressed to produce tablets with different
combinations of hardness, thickness, diameter, and weight to
determine the compression behavior of pellets and to charac-
terize the effect of tablet dimensions on drug-release properties.
Characterization. Prepared pellets were evaluated for particle-
size distribution, flow, and compressibility characteristics. In
addition, pellets were subjected to both light and scanning elec-
tron microscopy (SEM). True density of pellets was determined
using the solvent displacement method with benzene.
In vitro drug-release studies. Dissolution studies with DLT pellets
or tablets (equivalent to 90 mg DLT; n = 3–6) were conducted
using USP Apparatus I (rotating basket) dissolution method in
a programmable dissolution tester USP Apparatus XXI/XXII
(TDT-0P, Electrolab, India). Test specifications were as follows:
USP I baskets rotating at 100 rpm with 900 mL of distilled water
maintained at 37  0.5 C as dissolution medium. The 5-mL
samples were withdrawn at various time intervals during a 24-
h time period, and the volume was immediately replenished
with a fresh medium. Samples were filtered through a 0.45-m
pore size filter (Minisart, Sartorius AG, Germany) and analyzed
spectrophotometrically (DU640i, Beckman Coulter, Miami, FL)
at 237 nm. Pellet shape, and any visible changes in shape, struc-
www.phar mtech.com
 Table I: Differential compositions of different DLT pellet formulations using extrusion–spheronization.
Ingredient concentration (% w/w)
Granulation fluid
FC DLT MCC MS GMS GPS EC HPMC (weight used)* MC
D1 33 62 5 — — — — 5% w/w HPMC in water (28.44 g) 20.72
D2 33 59 8 — — — — 5% w/w HPMC in water (19.34 g) 18.71
D3 33 56 11 — — — — 5% w/w HPMC in water (20.40 g) 18.59
D4 33 57 — — 10 — — Water (15.0 g) 18.58
D5 33 47 — — 20 — — Water (11.5 g) 14.91
D6 33 47 — — 15 — 5 Water (13.0 g) 15.87
D7 33 47 — — 10 — 10 Water (16.5 g) 19.2
D8 33 52 — — — — 15 Water (17.5 g) 21.32
D11 33 47 — 20 — — — 5% w/w HPMC in water (20.71 g) 18.99
D12 33 17 — 50 — — — 5% w/w HPMC in water (6.61 g) 10.72
D13 33 32 — 35 — — — 10% w/w HPMC in water (6.68 g) 11.01
D14 33 37 — 30 — — — 30% w/w PVP in water (13.51 g) 9.0
D15 33 32 — 25 — 10 — 10% w/w HPMC in water (13.78 g) 13.50
D16 33 32 — 10 — 25 — 10% w/w HPMC in water (15.4 g) 16.32
D17 33 22 — 25 — 20 — 10% w/w HPMC in water (12.36 g) 13.39
D18 20 38 — 30 — 12 — 10% w/w HPMC in water (15.17 g) 16.77
D33 30 35 3 15 — — 5 Eudragit NE 30 D dispersion 19.30
(24.33 g) added in three parts to
produce 12% w/w solids
*Values in parentheses are the amount of granulation fluid used.
Abbreviations: FC is formulation code, DLT is diltiazem hydrochloride; MCC is microcrystalline cellulose; MS is magnesium stearate; GMS
is glyceryl monostearate; GPS is glyceryl palmitostearate; EC is ethylcellulose; HPMC is hydroxypropyl methylcellulose; and MC is
moisture content.

ture, and integrity during the dissolution study were recorded at
the end of the experiment. Pellet residue remaining after the dis-
solution studies was dried at room temperature and preserved for
SEM.
Release profiles comparison. Mathematical comparisons of disso-
lution curves from various formulations provide an opportu-
nity to test the similarity between two dissolution profiles. Fit
factors (f1 and f2) proposed by Moore and Flanner were used for
comparing the dissolution profiles (14). An f2 value of 50 in-
dicates similarity between two dissolution curves, whereas f1 is
used as an additional parameter to confirm the similarity when
the value is 15. In addition to fit factors, two model indepen-
dent ratio parameters—mean dissolution time (MDT) and dis-
solution efficiency (DE)—were calculated for the DLT formu-
lations for comparison of their drug-release profiles (15–17).
Release kinetics and mechanism. Drug-release data obtained from se-
lected formulations were subjected to various drug-release mod-
els to establish mechanisms as well as kinetics of drug release as
described previously (10). Criteria for selecting the most appro-
priate model were based on the highest coefficient of determina-
tion (R2), the smallest sum of squared residuals (SSQ), and Akaike
information criteria (AIC) (18). R2 was used to compare the fit of
a formulation with various kinetic models, whereas, AIC and SSQ
in combination with R2 values were used to compare the fit of
various formulations with the same kinetic model. F-statistics
were used to check the occurrence of correlations by chance.
Stability studies. A freshly prepared batch of optimized DLT for-
66 Pharmaceutical Technology APRIL 2004
mulation was taken in 30-mL screw-capped HDPE bottles and
stored in a refrigerator (Ultra, Godrej, India) at 8 C (control
sample) and in a stability chamber (WTC, Binder, Tuttlingen,
Germany) maintained at accelerated temperature and humid-
ity conditions of 40 C and 75% RH. After four months of stor-
age, samples from both stations were removed and analyzed for
visual characteristics, crushing strength, drug content, and drug-
release properties.
In vivo performance prediction of selected formulation. Drug-release pa-
rameters (R0 and tDEL) obtained from in vitro data and the drug’s
pharmacokinetic properties were used for predicting blood drug
concentrations–time profiles from single dose and at steady state
from multiple dosing (11). The method of superposition was used
for the steady-state concentration predictions. Values of Cssmax and
Cssmin, maximum and minimum steady-state concentration, re-
spectively, were compared with the desired values calculated from
a theoretically developed controlled drug-release profile.
The quality of CR formulations was evaluated by dosage form
index (DI) (19). In addition, the forgiveness indexes (FI) and
percent fluctuations were determined for the formulations (20).
FI determines how much latitude the patient has in delaying
the next dose, and the reduction in the post-dose duration of
action (Da) ensues inferior forgiveness (21).
Results and discussion
Desired drug-release profiles. Desired drug-release profiles for
twice-per-day DLT formulations were calculated from the drug’s
www.phar mtech.com
 Table I (continued): Differential compositions of different DLT pellet formulations using extrusion–spheronization.
Ingredient concentration (%w/w)
Granulation fluid
FC DLT MCC L-HPC 11 L-HPC 20 HPC GMS GPS BW (weight used)* MC**
D19 30.71 55.83 — — — — — — Eudragit NE 30 D dispersion 21.27
(24.11 g) to yield 13.46% w/w
solids
D20 30 40 — — — — — — Eudragit NE 30 D dispersion 23.14
(60g) to yield 30% w/w solids
D23 33 52 15 — — — — — Water (32.94 g) 36.2
D24 33 37 30 — — — — — Water (28.72 g) 39.61
D25 33 30 37 — — — — — 30% w/w PVP in water 33.98
(21.88 g)
D26 33 37 — 30 — — — — Water (41.93 g) 40.91
D27 33 52 — — 15 — — — 30% w/w PVP in water 24.32
(24.37 g)
D28 33 52 — — 15 — — — Absolute alcohol:water —
(20 mL)
D29† 33 50 — — — 17 — — 20% w/w PVP in water 18.73
(18.79 g)
D30† 25 40 — — — — 35 — 20% w/w PVP in water 16.16
(16.13 g)
D31† 22 33.3 — — — — — 44.7 20% w/w PVP in water 12.09
(10.23 g)
D32† 25 55 — — — — — 20 20% w/w PVP in water 20.71
(18.61 g)
*Values in parentheses are amount of granulation fluid used.
**Moisture content of wet mass (taken just before extrusion) determined as % w/w on wet weight basis. Batch size was 50 g.
†
Drug was dispersed in molten LHA and mixture was cooled, powdered, and used for ES.
Abbreviations: FC is formulation code; DLT is diltiazem hydrochloride; MCC is microcrystalline cellulose; L-HPC is low substituted
hydroxypropyl cellulose; GMS is glyceryl monostearate; GPS is glyceryl palmitostearate; BW is bees wax; and MC is moisture content.

pharmacokinetic characteristics and desired steady-state drug able limits. All strategies were derived from one or more of the
concentrations (50–150 ng/mL) (22, 23). The desired drug- following modifications in the formulation development
release parameters were 94-mg doses to be delivered at a first- (process and/or composition):
order kinetic rate of 0.3309/h for a period of 6.95 h. The de- ● use of release-controlling excipients such as low–melting point

sired in vitro drug-release profile and predicted steady-state hydrophobic agents, water-swellable polymers, and water-
concentrations for dosing intervals of 12 and 24 h are shown insoluble excipients
in Figures 1a and 1b. ● drug used as powder, dissolved in granulation fluid, or dis-

Drug–excipient compatibility. Stability and compatibility of DLT persed in a low–melting point hydrophobic agent
with four types of excipients—water-insoluble, water-soluble, ● use of various aqueous systems as binders for granulation

low–melting point hydrophobic, and water-swellable cellulose (water, aqueous solution of binders, or aqueous dispersion of
polymers—were tested using DSC and FTIR. Enthalpy values (
H film-forming agent)
normalized to weight) and the melting endotherm temperatures ● drying temperature in combination with low–melting point

were obtained using DSC. For DLT–excipient mixtures, most of hydrophobic agents
the cases were preserved and did not show any significant differ- ● compression of pellets to produce tablets.

ence for the total study period of 60 days at 40 C and 75% RH.
The exception was lactose, which was eliminated because it showed
potential interaction during thermal stress testing. Similarly, an
FTIR spectrum of DLT was preserved and no changes were de-
tected in any of the tested DLT–excipient mixtures after storage
at 40 C and 75% RH for 45 days. The results ruled out chemical
incompatibilities in tested drug–excipient mixtures.
Formulation development. Various formulation strategies were
adopted and tested with the objective of generating pellets with
release characteristics and physical properties within accept-
68 Pharmaceutical Technology APRIL 2004
These approaches and their effects on product quality, phys-
ical properties such as yield (total and product), crushing strength,
densities (bulk, tapped, and true), pellet size, and drug-release
properties are described later in this article. The strategies are
separated into two sections: formulation composition modifi-
cations and process variable modifications. Ingredient compo-
sitions of the formulations prepared and their corresponding
physical properties are presented in Tables I and II, respectively.
Formulation modification approaches. Use of low–melting point hy-
drophobic agents. Low–melting point hydrophobic agents such as
www.phar mtech.com
 Table II: Physical properties of pellet formulations prepared using extrusion-spheronization.
Total Pellet Bulk Tap True Mean
yield yield Crushing density density CI density pellet Size Parameter
FC (% w/w) (% w/w) strength (N) (g/mL) (g/mL) (%) HR (g/mL) (µm) studied
D1* 90.06 50.58 6.15  1.47 0.740 0.791 6.38 1.068 1.1685 653
D2 75.50 44.48 5.53  1.23 0.801 0.871 8.00 1.087 1.2757 733
D3 64.81 56.16 7.37  2.15 0.757 0.863 12.26 1.140 1.1581 624
D4 45.00 40.18 4.89  1.31 0.742 0.754 3.98 1.040 1.2879 645
D5 41.16 38.36 2.41  0.62 0.611 0.679 10.00 1.111 1.2259 555 Use of LHA
D11 70.13 60.88 5.63  0.99 0.734 0.765 4.00 1.042 1.2881 715
D12 47.18 43.55 3.53  0.97 0.644 0.678 5.00 1.053 1.1684 691
D13 43.85 36.50 3.59  0.90 0.656 0.698 6.00 1.064 1.2230 484
D14 47.34 44.96 4.47  0.91 0.660 0.702 6.00 1.064 1.2635 590
D23 58.66 57.24 5.51  0.74 0.597 0.625 4.54 1.047 1.2289 569
D24 53.52 44.46 5.83  1.02 0.493 0.504 2.27 1.023 1.3910 513
D25 64.86 62.24 6.79  1.64 0.562 0.583 3.63 1.037 1.2061 569
D26 55.28 52.10 7.50  1.27 0.482 0.500 3.70 1.038 1.2646 532 Use of WS
D27 — — — — — — — — — polymers
D28 56.58 50.68 5.61  1.95 0.585 0.631 7.31 1.078 1.3044 577
D8 48.92 45.80 15.62  2.96 0.710 0.772 8.00 1.087 1.2304 609
D6 50.90 39.34 4.14  1.03 0.703 0.756 7.00 1.075 1.1295 633 Use of LHA,
D7 45.46 41.48 7.61  1.55 0.653 0.714 8.52 1.093 1.2207 755 WS, and WI
D15 58.30 56.26 4.29  0.91 0.695 0.724 4.01 1.042 1.1274 656 excipients/
D16 56.48 53.50 5.54  1.18 0.701 0.723 3.00 1.031 1.1304 553 polymers in
D17 53.56 50.84 4.46  0.87 0.650 0.692 6.00 1.064 1.2142 545 combination
D18 58.10 56.46 4.91  0.94 0.697 0.726 3.99 1.041 1.2099 607
D33 66.08 42.72 5.03  0.81 0.688 0.761 9.67 1.107 1.2259 735
D19 47.38 48.80 5.21  1.66 0.698 0.764 8.57 1.093 1.2103 524 Use of film-
D20 64.28 61.56 6.79  1.31 0.668 0.709 5.89 1.063 1.2180 538 forming agen
for granulation
D29 67.98 33.34 5.82  1.16 0.698 0.732 4.651 1.049 1.1557 827 Dispersion of
D30 60.98 46.54 4.96  1.01 0.714 0.741 3.571 1.037 1.1165 721 drug in
D31 48.92 47.28 4.36  0.86 0.619 0.677 8.57 1.093 1.1110 562 molten LHA
D32 62.94 46.10 4.88  0.71 0.592 0.684 13.51 1.156 1.1027 705
*Batch size was 100 g.
Abbreviations: FC is formulation code; CI is Carr’s index; HR is Hausner ratio; LHA is low–melting point hydrophobic agent; WS is water
swell able; and WI is water insoluble.

magnesium stearate, glyceryl monostearate, and glyceryl palmi-
tostearate were used as formulation components to make the ma-
trix hydrophobic and retard the drug-release rate. The drug was
mixed with dry-powder excipients and granulated using an aque-
ous polymeric binder solution. Because DLT is a highly water-
soluble drug, granulation fluid and the moisture content of wet
mass were expected to be very critical to the quality of the pellets.
On increasing magnesium stearate concentration from 5%
to 11% w/w (formulation D1, D2, and D3), the amount of gran-
ulation fluid required for extrusion was decreased from 28.44
to 20.40 g. The moisture levels attained were highest in formu-
lations with the lowest magnesium stearate concentration. This
result was attributed to reduced concentrations of SPA in the
formulations (these formulations had MCC as the only
moisture-holding component) as magnesium stearate was in-
corporated at the expense of SPA. Total and pellet yield were
gradually reduced with the increase in magnesium stearate con-
centration. Spherical pellets could be generated only with 5%
70 Pharmaceutical Technology APRIL 2004
w/w magnesium stearate. At higher concentrations, only dumb
bell– or cylindrical rod–shaped pellets were produced. Mean pel-
let size increased from 653 to 733 µm and then decreased again
to 624 µm with 5, 8, and 11% w/w magnesium stearate, respec-
tively (see Table III). No uniform trend in mean pellet size and
size distribution existed with increasing magnesium stearate con-
centration. This result could be erroneous because for 8% and
11% w/w magnesium stearate levels, the product was a cylindri-
cal rod shape and the size determination with sieve analysis could
not have determined the actual dimensions and size distribution
of product. The poor shape of the pellets was also reflected in high
Carr index (CI) and Hausner ratio (HR) values, which increased
as the concentrations of magnesium stearate increased. There-
fore, it can be concluded that magnesium stearate had adverse ef-
fects on the spheronizing property of powdered mixtures. In ad-
dition, the poor aqueous affinity of magnesium stearate and its
addition to formulations imparted a poor molding capacity to
otherwise good spheronization characteristics of SPA.
www.phar mtech.com
 Table III: Drug-release profiles comparison of tablets prepared from pellets of various formulation compositions using
model-dependent (first-order kinetic model) and model-independent parameters. Mean dissolution time for
pellets is presented for comparison with corresponding tablets generated by compression of the pellets.
First-order kinetic model parameters Model-independent parameters
MDT
FC* R2 k1 SE SSQ F AIC f2 f1 DE12 h
TAB Pellet
Desired 1.000 0.3311 0.0002 0.000 6E+07 –198.9 100 0 76.31 2.84 2.84
D4t 0.6617 0.1317 0.3041 1.110 23.47 5.46 54.65 14.34 80.29 2.22 0.82
D5t 0.5325 0.1316 0.3982 1.903 13.67 13.00 42.18 23.86 85.68 1.38 0.25
D6t 0.6576 0.1177 0.2743 0.903 23.05 2.57 47.79 19.85 80.78 2.09 0.20
D7t 0.7445 0.1088 0.2057 0.508 34.96 –5.49 59.05 12.26 75.78 2.42 0.41
D8t 0.4326 0.0371 0.1372 0.226 9.15 –16.83 42.15 23.06 60.95 1.54 0.23
D9t 0.6501 0.0571 0.1353 0.220 22.29 –17.21 40.51 27.55 64.11 3.33 0.56
D10t 0.5306 0.0556 0.1688 0.342 13.57 –11.03 45.87 22.30 69.60 1.51 0.37
D13t 0.7954 0.0433 0.0709 0.060 46.64 –35.31 33.58 32.08 48.70 3.86 1.46
D15t 0.7406 0.0340 0.0650 0.051 34.27 –37.72 30.91 36.00 44.92 3.82 1.89
D16t 0.6852 0.0473 0.1036 0.129 26.12 –24.70 40.56 25.01 58.13 2.66 0.79
D17t 0.6949 0.0336 0.0719 0.062 27.33 –34.93 31.83 33.75 46.22 3.46 1.02
D19t 0.6264 0.0683 0.1703 0.348 20.12 –10.78 46.93 21.25 70.79 2.26 0.42
D20t 0.6224 0.0615 0.1547 0.287 19.78 –13.46 49.28 15.51 66.37 2.23 0.21
D21t 0.6646 0.0955 0.1812 0.576 23.78 –3.72 55.59 12.59 73.85 2.55 0.31
D22t 0.2689 0.0475 0.2529 0.768 4.41 0.30 51.73 15.70 70.50 1.95 0.35
D23t 0.6810 0.6978 0.2802 0.550 14.94 –4.38 27.84 44.99 33.74 0.72 0.19
D24t 0.5378 0.7639 0.4132 1.195 8.14 6.50 25.34 50.14 34.77 0.67 0.17
D25t 0.2229 0.0376 0.2269 0.618 3.44 –2.74 29.64 45.07 73.27 1.68 0.20
D26t 0.4025 0.0938 0.3984 1.588 6.74 10.47 15.17 86.17 91.32 0.67 0.17
D28t 0.4491 0.1021 0.3651 1.600 9.78 10.58 39.78 27.33 83.46 0.40 0.18
D29t 0.9777 0.2757 0.0700 0.049 439.3 –32.17 55.62 13.81 72.73 3.21 0.24
D30t 0.8384 0.0527 0.0746 0.067 62.25 –33.87 33.80 33.87 49.08 3.96 0.15
D31t 0.9792 0.0752 0.0354 0.015 563.6 –54.74 29.52 42.61 43.98 6.86 0.33
D32t 0.5170 0.0494 0.1543 0.286 12.84 –13.55 42.38 22.61 63.57 1.69 0.13
D33t 0.5256 0.0614 0.1883 0.425 13.30 –7.96 51.25 13.46 66.86 2.18 0.40
* “t” and the formulation code indicate that the tablet was prepared from the corresponding batch of pellet formulation.
Abbreviations: FC is formulation code; R2 is coefficient of determination; k1 is first-order drug release rate constant (hr–1); SE is standard error of
y–estimation; SSQ is sum of squared residuals; F is observed F value for F statistic; AIC is Akaike information criterion; MDT is mean dissolution
time (h); f2 and f1=similarity and dissimilarity factors, respectively, and were determined for different formulations by taking desired profile as
reference profile; DE12h is dissolution efficiency (%) with respect to 98.12% drug release in 12 h for twice-per-day formulations; and TAB is tablet.

To improve the physical characteristics, two other excipients
belonging to the same category, glyceryl monostearate and glyc-
eryl palmitostearate, also were tested. These excipients had bet-
ter affinities to water as reflected by their relative ease of wet-
ting. Consistent with the results produced with magnesium
stearate, the use of glyceryl monostearate also was character-
ized with lower moisture-level requirements of high concen-
trations. The moisture content of the wet mass required was
drastically reduced (see Table I), which also affects the pellet
yield adversely (see Table II). Unlike pellets made using mag-
nesium stearate, spherical pellets could be generated even at a
high concentration of glyceryl monostearate. Mean pellet size
decreased with an increase in glyceryl monostearate content be-
cause of a shift in size distribution in favor of the lower size
range, which was attributed to a lack of glyceryl monostearate
to provide sufficient cohesiveness to wet mass. This resulted in
a higher degree of fragmentation during spheronization. Poor
densification during spheronization also was reflected in a de-
72 Pharmaceutical Technology APRIL 2004
creased true density, bulk and tap density, and lower crushing
strengths of pellets. The decrease in glyceryl monostearate con-
tent from 50% to 35% w/w (D13) but a higher binder concen-
tration for granulation (HPMC as 10% w/w aqueous solution)
required a similar amount of granulation fluid and resulted in
decreased pellet yield (36% w/w). All other physical properties
of the pellets (D13) were similar to those of D12 except mean
pellet size. Excessive fines were produced as indicated by a de-
creased mean pellet size for D13. This result was caused by the
insufficient moisture content of wet mass that was required to
create desired levels of agglomeration during spheronization.
However, with 10% w/w concentration of HPMC in water, a
moisture level of 11% w/w in wet mass was sufficient for ES be-
cause the material had started becoming sticky and any further
addition of granulation fluid would have hampered the ES
process. D13 was modified by reducing GMS levels to 30% w/w
and replacing binder HPMC with PVP (30% w/w aqueous so-
lution) in granulation fluid (D14). This change not only im-
www.phar mtech.com

Figure 1: (a) Predicted steady-state blood drug concentrations for
twice-per-day DLT formulation. (b) Corresponding desired in vitro drug
release profile (first-order kinetics).
proved the pellet size, but also produced harder pellets. All other
physical properties were more or less unchanged.
Similar to glyceryl monostearate, mean pellet size, product
yield, and crushing strength decreased as the concentration of
glyceryl palmitostearate increased (see Table II). CI and HR val-
ues also increased which indicated adversely affected pellet shape
and flow characteristics. These results were attributed to re-
duced plasticity of wet mass, thereby leading to poor agglom-
eration and production of fines during spheronization.
Drug release from all these formulations containing magne-
sium stearate, glyceryl monostearate, and glyceryl palmitostearate
exhibited a high-burst release ( 40% within 1 h). The use of
various concentrations of low–melting point hydrophobic agents
in combination with various binders could not prevent the burst
release. In addition, high-burst release also was ascribed to high
aqueous solubility of drug, which led to the immediate release
of the drug available on the pellet surface. In addition, the high
surface area of the pellets further pronounced the burst effect
because the release of drug from the pellet’s surface enhances
the permeation of aqueous dissolution media into the pellet
matrix, which further results in rapid dissolution and diffusion
of drug out of the pellets. The use of a larger size fraction of
DLT pellets containing glyceryl palmitostearate did not improve
74 Pharmaceutical Technology APRIL 2004
Figure 2: Drug release profile from D29t (DLT-TAB) in comparison to the
drug release profile from D29 pellets (DLT-pellets) and desired drug
release profiles from a twice-per-day first-order kinetic CRDDS (Des-bid).
burst release, thereby indicating that high drug solubility was
the main reason for burst release.
Use of water-swellable polymers. Because the use of hydrophobic ex-
cipients did not produce the desired release profile, attempts were
made to control the drug release by using water-swellable poly-
mers as pellet matrix-forming agents. Water-swellable polymers
are commonly used as matrix-forming excipients to achieve CR
properties from drug delivery systems (24, 25). Upon contact
with aqueous media, these polymers swell to form a gel by means
of polymer chain relaxation mechanisms (26, 27). The drug re-
lease is controlled as a function of the swelling rate of the poly-
mer, the diffusion rate of aqueous media into the swelling ma-
trix, the diffusion rate of drug out of the swollen matrix, or the
erosion rate of swollen matrix depending on the characteristics
of the polymer used and the delivery system under study (24).
Four water-swellable polymers were incorporated as formula-
tion components at the expense of SPA. Various granulation
media—water, PVP in water, or an alcohol–water mixture—were
used to prepare a wet mass. In these preparations, DLT was dis-
solved and dispersed in small amounts of granulation fluid and
added to the dry-powder mixture comprising excipients. This
procedure was conducted to avoid overwetting during wet mass-
ing because the presence of a highly water-soluble drug and water-
swellable polymers in various grades and quantities were expected
to have tackiness problems, thereby leading to high unpredictability
in the amount of granulation fluid required for formulation. Sub-
sequent to the addition of drug solution–dispersion, extra gran-
ulation fluid was added to achieve the desired moisture content
in wet mass. The tendency of a powder mixture to form a tacky
wet mass was at a minimum with L-HPC 11 and was found to in-
crease with the use of higher viscosity grades of L-HPC 20 or poly-
mers with known higher swelling properties (HPMC).
Because water-swellable polymers have enormous capabili-
ties for holding water and swelling, a smaller amount of gran-
ulating fluid or water was required as the concentration of L-
HPC 11 increased from 15% (D23) to 30% (D24). Pellet yield
was reduced with a slight shift in size distribution in favor of a
lower size, thereby reducing mean pellet size from 569 to 513
µm (see Table II). Crushing strength was almost unchanged.
www.phar mtech.com
Pellets were of spherical shape and had
good flow properties as suggested by their
very low CI and HR values (see Table II).
However, pellets showed high disintegra-
tion immediately on coming in contact
with the aqueous medium in dissolution
studies, releasing almost 80% of the drug
within 15 min. Hence, a higher concen-
tration of L-HPC 11 was used in combi-
nation with the polymer binder solution
as aqueous granulation fluid (D25). Al-
though total and pellet yields were in-
creased in comparison with those of D23,
the proportions of pellets generated in the
desired size range were similar (see Table
II). Pellet hardness increased from 5.83 to
6.79 N (see Table II). This result was at-
tributed to the binder effect as well as to
a higher concentration of L-HPC 11 in the
D25 formulation. However, drug-release
properties were not improved, and simi-
lar to the release of D23 and D24, high
burst release was attained.
Because L-HPC 11 was ineffective in
avoiding the burst release effect, a higher
viscosity-grade polymer L-HPC 20 was
used (D26). To improve pellet strength,
the amount of granulation liquid was in-
creased, which was anticipated to avoid
immediate disintegration as observed with
L-HPC 11. Total yields and pellet yields
improved in comparison with those of
D25, which had higher L-HPC 11 in com-
bination with polymeric binder solution
used as granulating fluid (see Table II).
Though mean pellet size and crushing
strength increased slightly in comparison
with those of D25, the magnitude of burst
effect was unaltered.
Because higher viscosity or higher mol-
ecular weight grades of water-swellable
polymers form gels of higher viscosity and
higher gel stability compared with gels
formed by lower viscosity grades, a higher
viscosity grade of a water-swellable poly-
mer, hydroxypropyl cellulose was used in
the formulation (28). Upon granulating
the powder mixture with an aqueous poly-
meric binder, a high degree of tackiness
was observed with hydroxypropyl cellu-
lose. The batch (D27) could not be pre-
pared because the wet mass produced a
very sticky extrudate that clumped and
could not be collected. An attempt was
made with the ethanol–water mixture (1:1
v/v) to prepare a wet mass. With this com-
bination of solvents, pellet size and yield
were comparable with formulations con-
taining L-HPC 11 and hydroxypropyl cel-
lulose which could be generated but with
reduced hardness (see Table II). Reduced
crushing strength of pellets was caused by
the use of organic solvents in the granula-
tion liquid, because the use of organic liq-
uid as the granulation fluid reduces the
strength of pellets. Stronger pellets are pro-
duced as the mole fraction of water in a
water organic solvent mixture is increased
(29). Pellet density was higher than that gen-
erated for the L-HPC 20 formulation (D26),
thus indicating a higher degree of densifi-
cation achieved during spheronization.
From these results, it was concluded
that various grades and concentrations of
water-swellable polymers—L-HPC 11, L-
HPC 20, and hydroxypropyl cellulose—in
combination with various granulation flu-
ids could be used to produce pellets of ac-
ceptable quality and yields. However, the
expected CR property could not be
achieved. This finding was attributed to
the rapid swelling of water-swellable poly-
mers shown by all three grades to cause
disintegration of the pellets rather than
the formation of the gel structure. Hence,
HPMC, a different high molecular weight,
water-swellable polymer known to form
a swollen gel matrix at a slower rate upon
contact with aqueous fluids, was selected
to test the suitability of water-swellable
polymers to control drug release from a
matrix-based pellet system.
Wet mass preparation with HPMC
(D8) was extremely difficult because
HPMC showed a high tendency to form
a tacky mass. Smaller amounts of granu-
lation fluid (smaller than those required
for formulations containing L-HPC 11, L-
HPC 20, and hydroxypropyl cellulose) was
required to prepare the wet mass. Mean
pellet size increased because relatively
larger fractions of pellets were in the
710–1000 µm range. However, almost 93%
w/w of pellets were concentrated in the
355–1000 µm size range. Pellet yield was
relatively lower because of the loss of some
of the wet mass that stuck to the extruder
screen and spheronizer plate surface. The
crushing strength of the pellets was very
high (15 N), which was attributed to
HPMC, a high viscosity–grade polymer
that provides better cohesiveness even at
lower moisture content. The dissolution
studies showed that pellet disintegration
 integration, it was selected for use in combinations
with low melting hydrophobic agents. Similarly, eth-
ylcellulose, a commonly used matrix-forming agent
in tablets with excellent release retarding properties,
was selected for combinations with low–melting point
hydrophobic agents with the objective of retarding
burst release. Studies with combinations of these ex-
cipients are discussed in the following sections.
Use of glyceryl palmitostearate and HPMC. Glyceryl palmi-
tostearate and HPMC were combined in various
proportions keeping SPA and DLT levels constant
(D6 and D7). These results were compared with
those from D5 and D8, which included glyceryl
palmitostearate and HPMC alone, respectively. As
the ratio of the two components increased in favor
of HPMC, the amount of water required for wet
massing also increased. This observation was un-
like the results obtained with formulations con-
taining glyceryl palmitostearate (D4 and D5) or L-
HPC 11 (D23 and D24), in which moisture
requirements reduced with increasing proportions
of glyceryl palmitostearate or L-HPC 11 in the for-
mulation. The result was attributable to the de-
creasing content of SPA in these formulations (D4,
D5, D23, and D24) that was replaced with glyceryl
palmitostearate or L-HPC 11. All other physical
properties were similar to the formulations con-
taining glyceryl palmitostearate and L-HPC 11
alone. However, burst release was observed in both
formulations even at various glyceryl palmi-
tostearate and HPMC proportions.
Use of glyceryl monostearate and ethylcellulose. Glyceryl
monostearate and ethylcellulose were used in com-
bination along with an aqueous HPMC solution as
the granulation fluid. Glyceryl monostearate and
ethylcellulose were taken in various proportions
and studied. As the ratio of glyceryl monostearate
Figure 3a: Scanning electron micrographs of top (A–F) and radial fracture surfaces (G,
and ethylcellulose was changed from 2.5:1 (D15)
H) of tablets prepared from D29 pellets showing the effect of compression pressure on
to 1:2.5 w/w (D16) without changing the drug or
pellet deformation and fracture development. The mean hardness is 3.93 kPa for tablets
SPA content, the amount of granulation fluid re-
A–C and 6.77 kPa for tablets D–F, in increasing order of magnification (12, 37, and
quired for preparing the wet mass increased (see
170). G and H are tablets of 3.93 kPa and 6.77 kPa, respectively at 35 magnification.
Table I). This result was expected because hy-
drophobic glyceryl monostearate was replaced with
was delayed to 6–7 h. Although disintegration was not as rapid ethylcellulose, which has higher moisture holding capacity. Pel-
as those of the other water-swellable polymers, a burst release let yields were similar in both cases (54% w/w for D15 and
was still observed. Results suggested ineffectiveness of tested D16, respectively), but size distribution shifted in favor of the
water-swellable polymers, when used alone, in controlling the lower size which was reflected in the reduced mean pellet size
drug release from multiunit matrix-based particulate systems. (see Table II). However, drug-release properties were not im-
Use of excipients and polymers in combination. From the results ob- proved because burst release was still observed. Hence, ethyl-
tained using various low–melting point hydrophobic agents cellulose concentration further increased at the expense of SPA
and water-swellable polymers, it was concluded that good qual- to produce a glyceryl monostearate to ethylcellulose ratio of
ity pellets could be generated using these agents. However, burst 2.5:2 (D17). In comparison with D15, D17 required a slightly
release of DLT could not be controlled by means of matrix-based smaller amount of granulation fluid, which is attributed to a de-
pellets. Hence, combination strategies were tried, in which var- creased SPA content. There was no change in drug-release pro-
ious low–melting point hydrophobic agents were combined with files. Another formulation of D18 was prepared, in which the
water-insoluble excipients and water-swellable polymers. Be- glyceryl monostearate level was increased at the expense of DLT
cause pellets containing HPMC showed some resistence to dis- (decrease in drug content was replaced with SPA and glyceryl
76 Pharmaceutical Technology APRIL 2004 www.phar mtech.com
monostearate) keeping the ratio of glyceryl
monostearate to ethylcellulose similar to
that in D15 (i.e., 2.5:1 w/w). However, the
drug-release pattern was unaltered, and sig-
nificant burst release occurred.
An attempt was made to combine mag-
nesium stearate, glyceryl monostearate,
and ethylcellulose in one formulation with
PVP as the granulation fluid (D33). Al-
though pellet yield, mean pellet size, and
other physical properties were within the
acceptable range, drug release was char-
acterized by burst release (see Table II).
Use of film-forming polymeric dispersion.
An aqueous dispersion of film-forming
polymer was used as the granulating agent
because it was expected that the presence
of a film-forming polymer in a matrix
could help control the drug release. Solid
content of aqueous dispersion added to
the formulation was considered in the de-
termination of the percent composition
of various ingredients in the formulation
(D19). To increase the solids content in
the formulation, a higher amount of aque-
ous dispersion was used. The amount of
aqueous dispersion was divided into three
parts. The first part was mixed with a dry-
powder mixture to generate a wet mass
without forming lumps. The wet mass was
stored at 45 C in a forced-air oven for 30
min (complete drying was avoided), fol-
lowed by the addition of the second part
of the granulation fluid. The mass was
dried again under similar conditions and
the third part was added. The wet mass
thus produced was extruded and spher-
onized to generate pellets (D20). Pellet
yields were higher for D20 as the moisture
level attained was a slightly higher giving
fraction of pellets in the desired size range.
Mean pellet size was marginally increased
for D20 in comparison with that of pel-
lets from D19. Most of the pellets (95%
w/w) were concentrated in the 355–710
µm size range in both cases. The two for-
mulations reflected no improvement in
controlling the drug release.
Process modification approaches. Dispersion
of drug in molten LHA. Three low–melting point
hydrophobic agents—glyceryl mono-
stearate, glyceryl palmitostearate, and bees
wax—were used. The drug was dispersed
in the molten agents and the solid mass pro-
duced after cooling the mixture was pow-
dered to produce a powder of DLT and
low–melting point hydrophobic agents.
In the case of glyceryl monostearate
(D29) and glyceryl palmitostearate (D30),
DLT and low–melting point hydrophobic
agents were taken in a ratio of 1:0.5 and
1:1.4 w/w, respectively. A solidified mass
was crumbled by hand to yield powder.
For bees wax (D31 and D32), the drug-
to-bees wax ratio was 1:2 and 1:0.8 w/w,
respectively and the solidified mass needed
to be milled in a multimill to generate pow-
der. Powdered mass passed through BSS
no. 80 mesh was used for ES (see Table II).
As with other formulations, the moisture
requirement was controlled by SPA levels
so that lower SPA content in formulation
resulted in less granulation fluid required
to produce a wet mass. Pellet yields were
satisfactory (33–47% w/w). In the case of
bees wax, mean pellet size increased as the
DLT-to-bees wax ratio shifted from 1:2 to
1:0.8 w/w because of an increased SPA level
as bees wax was reduced. Despite good
physical properties, the drug release from
all these formulations could not be con-
trolled. However, the drug was dispersed in
a low–melting point hydrophobic agent but
breaking the solid mass into fine powder
may have generated free drug particles
rather than coated or embedded drug par-
ticles in the agent. Changing the agent or
using a higher low–melting point hy-
drophobic agent level did not improve
drug-release properties further.
Effect of pellet drying temperature. To exam-
ine the effect of drying treatment on drug-
release properties, the formulations con-
taining the low–melting point hydro-
phobic agent and film-forming materials
were dried at room and their melting tem-
peratures. Previous studies show that such
an approach can be used to yield matrix
pellets (29–31). It was hypothesized that
the drying of pellets at a slightly higher
temperature than the melting point of ma-
trix former would cause the molten ma-
trix former to fill the channels and capil-
laries of the matrix, thereby providing better
control over the drug release. Different dry-
ing treatments to generate pellets from var-
ious formulations did not produce any pos-
itive effect in retarding the drug release (data
not shown). Instead, an increased drug re-
lease was observed in all cases. This was at-
tributed to a disrupted matrix structure
caused by molten matrix former escaping
out of the pellets. The result was reflected
by the reduced true density of pellets caused

Figure 3b: Scanning electron micrographs of DLT-TAB showing top surface (A, 100),
radial fracture surface (B, 17), pellet aggregate obtained by radial fracture of tablet
(C, 40; E, 70 and F, 500), and single pellet obtained by radial fracture of tablet
(D, 60).
Figure 4: Drug-release profiles of DLT tablets (109- and 166-mg drug
content) prepared from two different size fractions of D29 pellets: 1000
µm (#16/166 mg and #16/109 mg) and 355–710 µm (#44/166 mg and
#44/109 mg).
by drying at an elevated temperature, suggesting an increased
porosity of pellets subsequent to drying treatment.
Compression of pellets: drug-release studies for comparative eval-
uation. On the basis of the results obtained from pellet formu-
78 Pharmaceutical Technology APRIL 2004
lations of DLT, it was concluded that neither the
formulation nor the process modification ap-
proaches were successful in achieving the desired
drug-release profiles. In all the formulations, a
prominent burst drug release was observed that
could not be controlled by any of the approaches.
Hence, pellets were compressed as an extension step
in the process of the preparation of a controlled-
release drug delivery system for DLT using ES. Pel-
lets (355–1000 µm size range) generated from var-
ious formulation compositions were compressed
on a single-stroke tablet compression machine to
produce tablets. Tablets were first evaluated for their
drug-release properties followed by other evalua-
tions for selected formulation. In general, as re-
flected by MDT, drug release from these tablets was
better controlled than corresponding drug-release
profiles produced from pellets (see Table III). Bet-
ter control over drug release from tablets was as-
cribed to close packing of the pellets on compres-
sion, thus forming a matrix structure with limited
porosity. Compression of pellets into single-unit
tablets also reduced the surface area of drug re-
leasing unit exposed to dissolution media, thereby
avoiding the immediate burst release observed in
the case of the pellets.
Comparative evaluation of various tablet formulations: for-
mulation selection. Various model-independent para-
meters and first-order kinetic model parameters
were determined from drug-release profiles for com-
paring the release properties of various formula-
tions (see Table III). Model-independent parame-
ters showed that D4t, D7t, D21t, D22t, D29t, and
D33t produced f2 values 50 when compared with
the desired profile. DE12 h (area under the dissolution curve up
to 12 h expressed as percentage of the area of rectangle described
by desired 98.12% release in 12 h) as well as MDT values for
these formulations were close to the desired values. Except for
D22t, the remaining formulations showed f1 values 15. These
results suggested similarity of drug-release profiles from five
tablet formulations to desired release profile. However, on fit-
ting the dissolution data in a first-order kinetic equation, D4t,
D7t, D21t, D22t, and D33t showed very poor correlation char-
acterized by R2, SSQ, AIC, and F-statistics. Only D29t fit well
in the first-order kinetic model and exhibited drug-release rate
(0.2757/h) close to the desired value (0.3311/h) with good cor-
relation (high R2, low SE, and SSQ, F-observed higher than F-
critical and low AIC). The drug-release profile for D29t is shown
in Figure 2. Hardness and friability of D29t also was determined.
Diametrical crushing strength of D29t was 16.60  0.46 kPa
and loss in the weight of the tablets upon friability testing was
0.5% w/w. Physical properties and yield of the pellet formu-
lation (D29) also were satisfactory for a given batch size (see
Table II). On the basis of these results, D29t was selected for the
development as first-order kinetic CR tablets of DLT. The se-
lected formulation was named DLT-TAB.
DLT-TAB. Compression behavior of pellets. D29 pellets were com-
www.phar mtech.com
Figure 5: Drug-release profile from halved and intact DLT-TAB tablets (109-mg drug content).
pressed to yield tablets of three hardness
values. Tablet weight was kept constant
(109- or 166-mg DLT content) and hard-
ness was varied using several compression
pressures to generate tablets of different
thicknesses. The mean hardness values
were 3.93, 6.77, and 16.60 kPa for 109-mg
DLT tablets and 3.03, 8.60, and 20.20 kPa
for 166-mg DLT tablets. The top and ra-
dial fracture surfaces of the tablets were
observed under scanning electron mi-
croscopy (SEM) to examine the frag-
mentation behavior of the pellets under
various compression pressures. The effect
of compression on drug-release proper-
ties also was studied. Photomicrographs
of 109-mg DLT tablets’ top and fracture
surfaces of various hardnesses were pre-
pared from D29 pellets (see Figures 3a and
3b). Tablets with low (3.93 kPa), medium
(6.77 kPa), and high (16.60 kPa) hardness
values were referred to as LP, MP, and HP,
respectively.
LP and MP tablets contained discrete
pellets with clearly distinguishable bound-
aries. The separation distance between the
pellets was smaller in the MP tablets, thus
indicating closer packing of pellets com-
pared with LP tablets. Although the pellets
kept their integrity without any signs of
fragmentation on compaction to generate
LP and MP tablets, deformation of the pel-
lets occured upon compression (Figure 3a).
In addition, fractures developed on the
pellets’ surfaces. Even though the fracture
depth was restricted to the outer periph-
ery of pellets, fractures were found pre-
dominantly at interpellet contact points
propagating toward the center of the pel-
lets (see Figure 3a). Pellet deformation in
the MP tablets was higher than in the LP
tablets. Upper as well as radial fracture
surface of the HP tablets (DLT-TAB) did
not show distinguishable pellet structures
(see Figure 3b; photomicrographs A and
B). However, pellets with clear boundaries
were visible in fragments generated by ra-
dial crushing of the tablets (see Figure 3b).
This finding indicates that the pellets ex-
isted in extremely close packing in HP
tablets. In a study of the compression be-
havior of microcrystalline cellulose pel-
lets, Johansson (32, 33) correlated tablet
tensile strength with tablet air permeabil-
ity and showed that lower permeability,
corresponding to a more-closed inter-
granular structure, produced tablets of
higher tensile strength.
Thus, a higher tensile strength of MP
tablets indicated a closed intergranular
pore structure. Pellet compression by de-
formation and a low incidence of frag-
mentation indicated that less energy was
required for repositioning the primary
particles in the pellets compared with the
energy needed to separate the primary
particles by formation of a fracture plane.
Johansson described the difference be-
tween energy requirements for shearing
and for fracturing resulting from of the
stress on individual pellets during uniax-
ial compression in a die (32). Simultane-
ous stress application on pellets from sev-
eral directions makes the fracturing of
pellets relatively difficult. Hence, com-
pression of the D29 pellets was caused pri-
 surfaces were exposed. Only halved tablets weigh-
ing 49% of the respective intact tablet weights
were used.
The results showed that breaking the DLT-TAB
tablets in half produced an adverse effect on the
drug-release properties (see Figure 5). Although the
f2 value was 51.55, which indicates a borderline sim-
ilarity of drug release, there was a distinctly faster
drug release observed in the halved tablets within
the 2–6-h period of drug release in this study. This
finding could be attributed to an increase in the ex-
posed surface area in the halved tablets. However,
drug release was similar to the halved and intact
tablets during the initial and terminal portions of
drug release. Absence of any burst release signified
that pellets were packed closely after compression,
thus preventing faster drug release in the early phase
of the dissolution study, despite the increase in the
halved tablets’ exposed surface area.
SEM pictures show that tablet structure relaxes
in an aqueous dissolution medium, thereby loos-
ening the pellets and creating channels for drug re-
lease (see Figure 6). Distinct pellet boundaries
within the tablet structures appear after exposure
to dissolution media. Hence, after the initial relax-
ation of the tablet structure in a dissolution
medium, a fast drug release was observed. In the later
stages of time duration, the drug release leveled off
Figure 6: Scanning electron micrographs of DLT-TAB after 24-h dissolution study.
with the release profile from intact tablets. A similar
Tablets A–C are top surface views in increasing order of magnification (15, 60, and
effect would also have been operative with intact DLT-
500); D–F are radial fracture surface views in increasing order of magnifications (18,
TAB tablets, however, the characteristic fast drug re-
65, and 500).
lease in the middle part of the drug-release profile
was expected to be less prominent with intact tablets
marily by deformation such that the pellets remained coherent because of the smaller drug-releasing surface area.
units without significant flaws in their structures after Drug-release kinetics and mechanism evaluation. DLT-TAB tablets re-
compression. mained intact after a 24-h drug-release time period without los-
Effect of pellet size. To study the effect of pellet size on the drug- ing their shapes or sizes. However, cracks were observed on the
release properties of the tablets, different size fractions of D29 surface of the tablets during the drug-release studies. From SEM
pellets (retained on BSS no. 16, i.e., 1000 µm and 22/44 sieve, studies, the boundaries of the pellets were clearly visible after the
i.e, 355–710 µm) were used to prepare 109- and 166-mg DLT dissolution study on the outer tablet and the radial tablet frac-
tablets. ture surfaces, thus indicating that the pellets were intact after
The drug release was similar for tablets prepared from two compression. This finding is attributable to the closely packed
size fractions at both the tested dose levels of DLT, which is in- pellets after compression under high pressure, which resulted in
dicated by f2 values 50 (see Figure 4). The similarity between tablets without demarcated pellet boundaries. However, during
the two size fractions of pellets in terms of drug-release prop- drug-release studies, the tablet structure relaxes to loosen the
erties suggest that they could be mixed and used for tablet prepa- whole matrix. As a result, the individual pellet boundary is shown
ration, thus increasing the usable pellet yield. Upon correlating in the tablet structure remaining after the drug-release studies
this finding with SEM results, it was reaffirmed that even though were conducted (see Figure 6). Structural integrity as well as the
discrete pellets within tablets were the actual drug-releasing retained shapes and sizes of the tablets during the drug-release
units, the drug-release retardation effect was a function of the study indicated that the drug-release mechanism was diffusion-
close packing of pellets irrespective of the pellet size used in the and/or dissolution-controlled without any significant contribu-
tablet preparation. tion of erosion. However, the controlled-release dissolution of
Effect of tablet structural integrity. The drug-release property of the the highly water-soluble drugs in an aqueous media was less prob-
DLT-TAB formulation was studied as a function of the struc- able, thereby leaving diffusion to be the drug-release mechanism.
tural integrity of the tablet. DLT-TAB was halved by diametri- Drug-release data from the DLT-TAB tablets study were ap-
cally compressing with a tablet hardness tester (TBH 20, Er- plied to various kinetic models to elucidate the mechanisms
weka) at 2.3 mm/s until it fractured and the radial fracture and kinetics of drug release. The data fit well in the Baker–
80 Pharmaceutical Technology APRIL 2004 www.phar mtech.com
 Table IV: Kinetic and statistical parameters generated when applying the drug-release data of DLT-TAB tablets to various
mathematical models.
Zero First Hixson– Baker–
Parameter order order Higuchi Crowell Lonsdale Hopfenberg Weibull Peppas
Parameters for assessing fit of model
r 0.9187 –0.9888 0.9896 0.9896 0.9943 0.9786 0.9824 0.9925
R2 0.8439 0.9777 0.9793 0.9794 0.9886 0.9576 0.9651 0.9851
F 54.08 439.33 473.05 474.34 864.5 225.65 249.16 593.93
SSQ 1478.61 0.0491 196.14 0.2024 0.0018 0.0281 0.051 0.0082
AIC 91.59 –32.17 67.35 –15.17 –71.64 –38.86 –28.74 –48.84

Parameters for kinetic equation

Slope 7.29 –0.1197 28.481 0.2527 0.0325 0.0650 b = 0.7125 n = 0.441
a = 2.209
Td = 3.041
k 7.95 0.2757 28.481 0.2527 0.0325 0.0650 — 0.3482
SE (slope) 0.9916 0.0057 1.3094 0.0116 0.0011 0.0043 0.0451 0.0181
SE (y-est) 12.16 0.0700 4. 4288 0.1423 0.0136 0.0530 0.0753 0.0301
t12h — 96.34 — — — — 93.00 104.17
t90% — 8.35 — — — — 9.73 8.61
Abbreviations: r is correlation coefficient; R 2 is coefficient of determination; F is observed F–statistic value; SSQ is the sum of the squared
residuals; AIC is Akaike information criterion; k is the constant release rate with units of %/h, %/(h)1/2, (%)/h1/3 for zero-order, Higuchi, and
Hixson–Crowell models, respectively, and units of k for first order, Baker–Lonsdale, and Hopfenberg models are h–1. In the Peppas model,
units of k are hr–n; t90% is time taken for 90% drug release; SE (slope) is standard error of slope; SE (y–est) is standard error of y estimation;
and t12h is percentage of drug released in 12 h.

drug-releasing units were discrete pellets. This result also was

Table V: Analytical results of stability studies with indicated by the close fit of the dissolution data to the
DLT-TAB tablets stored at accelerated temperature and Baker–Lonsdale model, which is developed from the Higuchi
humidty test conditions (40 C and 75% RH) for 4 months. model and describes the drug release from spherical matrices.
After 4 months Hence, release retardation from the tablets, in comparison with
Estimated parameter Start of study of study the release from the pellets, was attributed primarily to the close
Drug content (% w/w packing of the pellets, which acted effectively to retard the pen-
(of label content) 97.41  0.64 94.96  1.83 etration of dissolution media into the tablets and consequently
First-order drug to prolong the drug release. At the same time, the reduced sur-
release rate (h–1) 0.2762 0.2770 face area of the pellets (upon compression into tablets) exposed
Similarity factor, f2 — 60.68 the dissolution media and prevented an immediate-burst re-
Dissimilarity factor, f1 — 8.96 lease. The tablets’ poor fit to the Hopfenberg model as well as
the unaltered tablet shape and size during the dissolution study
Hardness (kp) 16.60  0.46 13.73  1.29
eliminated the possibility of erosion-controlled drug release.
Visual inspection Smooth texture No visible
However, the high R2 value generated using the Hixson–
and good gloss changes
Crowell model was unexplainable because neither a diminished
tablet dimension was observed nor a drug release was suspected
Lonsdale, Higuchi, and first-order kinetic models (see Table to be limited by the drug’s particle dissolution rate.
IV). The compatible fit of the first-order kinetic and Higuchi To characterize the diffusion kinetics, data were applied to
models indicated that the drug release is controlled by a the Peppas model and the diffusional exponent n was deter-
concentration-dependent diffusion mechanism. This finding mined. The diffusion exponent n was 0.44, thereby indicating
indicated that the tablets were a heterogenous matrix system in the mode of drug release to be Fickian diffusion (see Table IV)
which the drug release occurred by means of a diffusion process (34, 35). Pharmaceutical systems that show a good fit to the
through channels or capillaries in the matrix. Peppas model also follow the Weibull model. DLT-TAB tablets’
However, SEM studies indicated that pellets did not fragment drug-release data were plotted according to the Weibull equa-
under compression to produce a matrix system. Instead, pel- tion to characterize the shape of the curve. Because the shape
lets were closely packed with some deformation. Although the parameter 1, the curve shape was expected to be parabolic
drug release occurred through the channels and pores present with a higher initial slope and consistent with the exponential
in between the closely packed pellets in the tablets by means of profile (see Table IV) (36).
diffusion (in agreement with Higuchian kinetics), the ultimate Accelerated stability studies. Stability studies for DLT-TAB tablets
82 Pharmaceutical Technology APRIL 2004 www.phar mtech.com
 the pellet surface, which con-
Table VI: Predicted steady-state plasma concentration levels of DLT from an in vitro tributed to a pronounced burst
drug-release profile of DLT-TAB tablets and the desired drug-release profile. effect. The drug release from the
Release tdel kr Dose Cssmin–Cssmax Dl DLT-TAB tablets took place by
profile (h) (h–1) (mg) (ng/mL) (% fluctuation) Fl concentration-dependent diffu-
Desired 6.95 0.3309 95 79.4–151.2 1.904 (62.77) 1.222 sion process through channels or
DLT-TAB 8.35 0.2757 95 83.2–144.2 1.733 (53.65) 1.245 capillaries in a pellet matrix and
Abbreviations: tdel is time of delivery; kr is drug release rate; DI is dosage form index (37); % fluctuation those in between closely packed
(20); FI is forgiveness index and is calculated from the predicted steady-state blood drug concentration pellets. The drug release followed
time profile with the assumption that the duration of action is correlated to blood drug concentration first-order kinetics and was con-
and is the period of dosing interval during which concentration remains above Cssmin (21). trolled by the degree to which the
pellets were closely packed. Re-
were conducted under accelerated temperature and humidity sults indicated that the DLT-TAB formulation was stable under
conditions (40 C and 75% RH) for 4 months. Results of the accelerated temperature and humidity conditions (40 oC and
drug assay, drug-release profiles, and hardness are presented in 75% RH) for 4 months. The in vitro drug-release profile from
Table V. DLT-TAB tablets was predicted to produce steady-state drug con-
No significant change was observed in the drug content of centrations within a desired range.
the tablets and was comparable with the drug content of the
control samples. There were no signs of distinguishable changes Acknowledgements
in appearance, color, and texture after 4 months of storage under The authors thank the Council for Scientific and Industrial Re-
accelerated conditions. The drug-release profiles were similar search, New Delhi, India, for its senior research fellowship.
because f2 values were 50 for the control versus the test sam-
ples (see Table V). DLT-TAB tablets were stable for 4 months References
under accelerated stability test conditions.
Prediction of in vivo concentration–time profile from in vitro data. A steady-
state drug concentration–time profile for the DLT-TAB tablets
was predicted using its in vitro first-order kinetic release rate.
Because the drug-release rate from the DLT-TAB tablets was
slightly lower than the desired level (see Table VI), the predicted
drug concentrations were also proportionately lower to the de-
sired concentrations. DI and FI were calculated (see Table VI).
Steady-state concentration predictions indicated that DLT-TAB
tablets generated slightly lower FI values than those predicted
from the desired drug-release profile because of a slower drug
release from DLT-TAB tablets (see Table VI).
Conclusion
Extrusion–spheronization (ES), though an easy method of mak-
ing multiunit particulate systems, offers very limited applicabil-
ity because of the limited number of excipients that can be used
for successful ES. Because the concept of matrix-based systems
(single or multiunit) for controlled drug delivery is expected to
provide time and cost advantages over other coating based sys-
tems, work was undertaken with the objective of identifying the
feasibility of using ES for the development of matrix-based,
controlled-release (CR) formulations for diltiazem hydrochlo-
ride (DLT), a highly water-soluble BCS Class I drug. A matrix-
based multiunit pellet system for CR delivery of DLT could not
be developed through various formulation and process modifi-
cation strategies adopted in the present study. The high aqueous
solubility of the drug and the large surface area of the pellets re-
sulted in extensive burst release. Although pellet systems were
characterized with high burst release, DLT tablets prepared by
compression of the pellets gave the desired drug-release proper-
ties. In addition, the medium dose requirements made it neces-
sary to incorporate drug in a higher proportion in the delivery
system. This resulted in a higher fraction of drug available on
84 Pharmaceutical Technology APRIL 2004
1. P. DeHaan and C.F. Lerk, “Oral Controlled Release Dosage Forms: A
Review,” Pharm. Weekbl. Sci. Ed. 6, 57–67 (1984).
2. V.H.K. Li and V.H.L. Lee, “Influence of Drug Properties and Routes
of Drug Administration on the Design of Sustained and Controlled
Release Systems,” in Controlled Drug Delivery: Fundamentals and Ap-
plications, J.R Robinson and V.H.L. Lee, Eds. (Marcel Dekker, Inc., New
York, NY 1987), pp. 3–94.
3. J.R. Cardinal, “Matrix Systems,” in Medical Applications of Controlled
Release: Classes of Systems, R.S. Langer and D.L. Wise, Eds. (CRC Press,
Boca Raton, FL, 1984), pp. 41–67.
4. D.A. Reynolds, “A New Technique for the Production of Spherical Par-
ticles,” Mfg. Chem. Aerosol News, 41, 40–43 (1970).
5. J.W. Conine and H.R. Hadley, “Preparation of Small Solid Pharma-
ceutical Spheres,” Drug Cosmet. Ind. 106, 38–41(1970).
6. C. Tapia, G. Buckton, and J.M. Newton, “Factors Influencing the Mech-
anism of Release from Sustained Release Matrix Pellets produced by
Extrusion/Spheronization,” Int. J. Pharm. 92 (2), 211–218 (1993).
7. S.R. Goskonda, G.A. Hileman, and S.M. Upadrashta, “Controlled Re-
lease Pellets by Extrusion–Spheronization,” Int. J. Pharm. 111 (1),
89–97 (1994).
8. D. Blanque et al., “Some Factors Influencing the Formation and In
Vitro Drug Release from Matrix Pellets Prepared by Extrusion/Spher-
onization,” Int. J. Pharm. 119 (2), 203–211 (1995).
9. S.H. Neau, M.Y. Chow, and M.J. Durrani, “Fabrication and Charac-
terization of Extruded and Spheronized Beads Containing Carbopol
974P, NF Resin,” Int. J. Pharm. 131 (1), 47–55 (1996).
10. A. Sood and R. Panchagnula, “Drug Release Evaluation of Diltiazem
CR Preparations,” Int. J. Pharm. 175 (1), 95–107 (1998).
11. W.A. Ritschel, “Biopharmaceutic and Pharmacokinetic Aspects in the
Design of Controlled Release Peroral Drug Delivery Systems,” Drug
Dev. Ind. Pharm. 15 (6, 7), 1073–1103 (1989).
12. R. Gandhi, C.L. Kaul, and R. Panchagnula, “Extrusion and Spheroni-
sation in the Development of Oral Controlled Release Dosage Forms,”
Pharm. Sci. Technol. Today, 2 (4), 160–170 (1999).
13. P. Kleinebudde, “The Crystallite–Gel Model for Microcrystalline Cel-
lulose in Wet Granulation, Extrusion, and Spheronization,” Pharm.
Res. 14 (6), 804–809 (1997).
14. J.W. Moore and H.H. Flanner, “Mathematical Comparison of Disso-
lution Profiles,” Pharm. Technol. 20 (6), 64–74 (1996).
15. K.A. Khan and C.T. Rhodes, “Effect of Compression Pressure on the
Dissolution Efficiency of Some Direct Compression Systems,” Pharm.
www.phar mtech.com
 Acta. Helv. 47, 594–607 (1972).
16. K.A. Khan, “The Concept of Dissolution Efficiency,” J. Pharm. Phar-
macol. 27, 48–49 (1975).
17. V. Pillay and R. Fassihi, “Evaluation and Comparison of Dissolution
Data Derived from Different Modified Release Dosage Forms: An Al-
ternative Method,” J. Controlled Release 55 (1), 45–55 (1998).
18. P.V. Parab, C.K. Oh, and W.A. Ritschel, “Sustained Release from Pre-
cirolä (Glycerol Palmitostearate) Matrix: Effect of Mannitol and Hy-
droxypropyl Methylcellulose on the Release of Theophylline,” Drug
Dev. Ind. Pharm. 12 (8, 9), 1309–1327 (1986).
19. M. Gibaldi and D. Perrier, in Pharmacokinetics (Marcel Dekker, Inc.,
New York, NY, 1982), pp. 182.
20. J.P. Skelly and W.H. Barr, “Regulatory Assessment,” in Controlled Drug
Delivery: Fundamentals and Applications, J.R. Robinson and V.H.L.
Lee, Eds. (Marcel Dekker, Inc., New York, NY, 1987), pp. 293–334.
21. J. Urquhart, “How Much Compliance Is Enough?” Pharm. Res. 13 (1),
10–11 (1996).
22. L.Z. Benet, S. Oie, and J.B. Schwartz,“Design and Optimization of Dosage
Regimens: Pharmacokinetic Data,” in Goodman and Gilman's The Phar-
macological Basis of Therapeutics, J.G. Hardman and L.E. Limbird, Eds.
(McGraw-Hill, New York, NY 1996), pp. 1707–1792.
23. W.A. Ritschel and G.L. Kearns, “Pharmacokinetic Parameters of Im-
portant Drugs,” in Handbook of Basic Pharmacokinetics (American Phar-
maceutical Association, Washington, DC, 1999), pp. 479–503.
24. K.V.R. Rao and K.P. Devi,“Swelling Controlled-Release Systems: Recent
Developments and Applications,” Int. J. Pharm. 48 (1–3), 1–13 (1988).
25. K.E. Uhrich et al., “Polymeric Systems for Controlled Drug Release,”
Chem. Rev. 99 (11), 3181–3198 (1999).
26. A.R. Rajabi-Siahboomi et al.,“Imaging the Internal Structure of the Gel
Layer in Hydrophilic Matrix Systems by NMR Microscopy,” J. Pharm.
Pharmacol. 44 (supp), 1062 (1992).
27. P. Colombo,“Swelling-Controlled Release in Hydrogel Matrices for Oral
Route,” Adv. Drug Del. Rev. 11 (1–2), 37–57 (1993).
28. R. Chatlapalli and B.D. Rohera, “Physical Characterization of HPMC
and HEC and Investigation of Their Use as Pelletization Aids,” Int. J.
Pharm. 161 (2), 179–193 (1998).
29. E.S. Ghali, G.H. Klinger, and J.B. Schwartz,“Thermal Treatment of Beads
with Wax for Controlled Release,” Drug Dev. Ind. Pharm. 15 (9),
1311–1328 (1989).
30. K.K. Peh and K.H. Yuen, “Development and In Vitro Evaluation of a
Novel Multiparticulate Matrix Controlled Release Formulation of
Theophylline,” Drug Dev. Ind. Pharm. 21 (13), 1545–1555 (1995).
31. Y. Miyagawa et al., “Controlled-Release of Diclofenac Sodium from
Wax Matrix Granule,” Int. J. Pharm. 138 (2), 215–224 (1996).
32. B. Johansson et al., “Compression Behaviour and Compactability of
Microcrystalline Cellulose Pellets in Relationship to Their Pore Struc-
ture and Mechanical Properties,” Int. J. Pharm. 117 (1), 57–73 (1995).
33. B. Johansson and G. Alderborn, “Degree of Pellet Deformation during
Compaction and Its Relationship to the Tensile Strength of Tablets Formed
of Microcrystalline Cellulose Pellets,” Int. J. Pharm. 132 (1, 2), 207–220
(1996).
34. N.A. Peppas, “Analysis of Fickian and non-Fickian Drug Release From
Polymers,” Pharm. Acta. Helv. 60, 110–111 (1985).
35. A. Sood and R. Panchagnula,“Role of Dissolution Studies in Controlled
Release Drug Delivery Systems,” S.T.P. Pharm. Sci. 9, 157–168 (1999).
36. F. Langenbucher,“Linearization of Dissolution Rate Curves by the Weibull
Distribution,” J. Pharm. Pharmacol. 24, 979–981 (1972).
37. F. Theeuwes and W. Bayne,“Dosage Form Index: An Objective Criterion
for Evaluation and Controlled-Release Drug Delivery Systems,” J. Pharm.
Sci. 66 (10), 1388–1392 (1977). PT

Please rate this article.

On the Reader Service Card, circle a number:
342 Very useful and informative
343 Somewhat useful and informative
344 Not useful or informative
Your feedback is important to us.

Pharmaceutical Technology APRIL 2004 85